MANGO Toolbox: Mitigating Atmospheric Noise with GNSS Observations

From repeat-pass interferometry (dInSAR), tropospheric signals often prevent the detection of ground deformation signals on active volcanoes. In past years, different tropospheric corrections have been implemented in InSAR automated processing systems based either on empirical methods or global weather-based models. However, these models face key challenges: limited spatial resolution (>10 km) and significant time latency (several days) for data availability. Local GNSS networks offer a promising alternative, delivering real-time tropospheric delay data, yet their potential in dInSAR corrections remains underutilized. In this study, we introduce MANGO (Mitigating Atmospheric Noise with GNSS Observations) a Python toolbox designed to produce phase delay maps from raw GNSS Zenith Tropospheric Delays (ZTD) for correcting individual interferograms. First, we evaluate the performance of GNSS-based tropospheric corrections on two tropical volcanoes: Piton de la Fournaise and Merapi. Then, we compare our approach to the corrections obtained from global ECMWF (ERA5 and GACOS). Our results demonstrate that for Piton de la Fournaise, GNSS-based corrections (~34 GNSS stations) reduce noise in 90% of processed interferograms, outperforming ERA5 and GACOS corrections by 25% and 50%, respectively. For Merapi, the performance of GNSS-based corrections with only 5 stations reaches the same level as ERA5 corrections. After correcting individual interferograms, GNSS-based corrections increase the signal-to-noise ratio in InSAR time series allowing the detection of slow inter-eruptive signals at Piton de la Fournaise. Here, we show that GNSS-based models are an efficient alternative for the production of corrected InSAR time series. These products will be valuable for Volcano Observatories for supporting the ground monitoring of volcanic unrest.